RESIDUAL STRESSES IN IRON CASTINGS

Received w March 26, 1977: Revised on Aususl 9, 1977

fit thj.ypaperan attempt has been made lo mview criticnlly the metlzods cnzployed
iui the measurement of residual stresses, the teJt pieces used, causes of residual
:lresses in =dings, mechanism of residuui stress formation and relief of residual
dresser. ~t is concluded that the causes of residual stress formation are fairly well
:otdmtood,but the mechanism behind the stre.vs relief produced by different methods
is not as well understood. It is suggested that a detailed study of the various stress
,&f m&ds, independently or in combinafiun, is essential to ~n?dcrstandthc
niechanism.
Key words: Residual stress, iron castings.

The presence of locked-in residual stresses in cestings have been recog-

nised long b ~ c k . Castings with appueciable residual stresses are found
u, distort during storage, transportation, .machining and service. Many
examples of cracks occurring due Lo high stresses ti211 be fourld in the litera-
ture.' The residual stresses limit the external Ioad to which the cast
component can be subjected to. This is so because wherever the residual
btresses and stresses due to the exterilal load are in the same direction, the
ioad carrying ability of the component is only the difference between its
strength (tensile or compressive as the case may be) and the residual stless.
However, the presence of residual stresses is not always bad. From
the argument presented above, it is immediately obvious that residual
stresses are desirable if they counteract stresses arisin& from external forces.
m y instances residual stresses are purposely introduced to increase
the service l i e of the comp0ner.t. I t is a common practice nowadays to
* Materials Science Division, National Aeronautical Laboratory, Rangalore 560 017.
** Professor, ~epartment of Mechanical Enginwring, Indian Institute of Science,
Bat,llae 5x50 012.
+ Head, Materials Science Division, Nationd Aeronutical Laboratory, Bangalore 560 017.
h c i a t e Professor, Department of Mechanical Engineering, Indian Institute of
BsnBal0r-z 560 012.
323
Ilso-9
3 24 C. BALASINGH et al.
induce surface compressive residual stresses to enhafice the faiig~i $
compopcnts. Portevin and Pomey2 and Patterson and Dietzel3 carried
out detailed investigations to bring out the beneficial cffects of re$&
stress. They concluded that surface compressive stresses increaed tg
tensile st~ength,bending strength, impact strength, torsional strength and
hardness. Notch effect was also reduced. .

In majority of cases, especially when. thc castings are complex, it is

difficult to predict the distribution of residual stresses. In such cases
standard mcthod is to eliminate the residual stresses altogether. There
are two ways to achieve this : one is to make castings in the stress-free
condition by adjusting different variables. This requires understanding asto
how and why stresses arise, and what factors influence the magnitude of
the residual stresses. The second alternative is to produce castings wit6
stresses, and then subject them to treatments to reljeve the residual strem,

The methods employed for the measurement of residual stresses may h.

classified i ~ t omechanical, physical and chemical. The choice of a par&
cular method depends on the shape of tbe specimen, precjsion required 3nd
the location a~.ddirection along which stresses are Lo be weamred.
1. The four most important and vely widely used methods cornmg
under the category of mechmical methods are:
(i) Parting-out method
(ii) Lzyer removal method
( i The boring and turning method and
(iv) The hole drilling method.
h b u 1 3 - ~literature
t is available on these methods. In the report SAE
J 9364 (and also TR 1445) publishcd by the Society of Automative
neers, the details of different Lechniques are given cleZrly. Refin-
in the mathematical approach, and improvements in techniques are beiog
reported periodically. In the present article, only a brief mention of fEz
principle and the advantage of each method will be made.
(i) Parting-out method
'
The parting-out method can be used either as a first step in a de-
stress determimtion (using other methods) or as a method complete
itself. Thc method requires removal of a coupon from the perent in
 Residual Stresses in Iron Castings 325
be region of interest. The coupon can. have the shape of flat plate or a
or curved beam. Strain and/or curvature, produced in the parted
,t coupon, are measured to compute the stresses. . .
If the distribution of stress through the thickness of the coupon is
hornor can be reasonably assumed, the residual stress in the parent part
my be determined. If the principal stress directions are known, the
becomes simpler. One beam oriented in the direction of principal
stress is parted out and analysed. If the directions are not known, two
beams m y be necessary. However, since cutting in one direction will
&st the stresses in the other directions also, appropriate correction factors
are to be introduced into the calculations. The minimum size of the beam
removed should be 1 in. long. In case of plate it should be 1 in. square.
The method is ideally suited to determine average stresses in large part.
The error in measurement can. be quite large if steep stress gradients are
present.

(ii) Layer removal method

This is a very useful and sensitive method in determining the resi-
dual stresses in flat plates or beams in which the stress varies with the thick-
ness, particularly if the stress grzdient is sharp. The stress m a layer
is determined by removing the layer and measuring the strain and curva-
ture changes it produces on the remainder of the specimen. Strain gauges
are used for strain measurement and predsion curvature gauges for the
curvature measurement.
Phte coupons are used if the principal stress directions are u ~ ~ o W .
If the directions are known, beams parted out in the principal directions are
&. The thickness of the specimen needed depends on the extent of
s b s measurement required. In case only surface stresses are desired,
thiEkRess of the order of 118 in. is sufficient. If the en.tire stress distri-
bution is needed, the coupon s h o ~ l dbe much thicker. The determination
of sruface stresses requires only primary cuts, whereas for the evaluation
of entire stress distribution, primary and secondary cuts should be used.
thickness of the laye~sto be removed depends on various factors such
8s size of the specimen, precision of the measurement required, and steep-
LP:a Of the stress gradient.
If precise results are to be obtained, the machin.in.g technique should
bechosen such that the induced machining stresses have the minimum effect.
'Fk Patal techniques are grinding, etcmg or elect~opolishing. The layer
removal method is relatively complex, especially for tb.ick curved beams,
Computer programming is necessary if if~veral~peciI.Uensare to be studif&
There r.re experimental set-ups wherein c o ~ t i n u omeasurement
~~ of strain
and curvatwe, as electi-opolishing goes on, is possible and a computer
coupled to it .calculates the stresses.

(i) The boring and turning method

This method is applicable only to circular cylindrical members. l-he
tangentiel, radial a ~ longitudinal
d stress distributions along a rad,ius may
determined by removing layers from the bore or the outer surface snd
measuring the Ioqgitudinzl ax! tangentiel strain changes. The two basic
assumptions mzde are : (a) The stiesses are symmetrical about tlie central
axis a17.d constant :Iang its length ; (6) the removal of a layer produces a
constant change in the longitudinal stress at all points in the specimen.
The strains are measured on the side opposite to the one on which
the layers are removed and are a measure of the stresses released in the
removed layers. Lf complete stress distribution is desired, or if the stresses
at one surface (bore or outer surface) are to be determined with the grebtesl
possibl.: precision then primary and secondary layer removal may be used.
Foi example, suppose the stresses at th.e ssuface of a solid shaft 2re &slred.
Ther the shaft is bored f i ~ s t@rimmy cuts). This reduces the sti&.ess so
th2t secordary cuts at the surface produce signilicant strain changes. Simi-
larly in the cese of r thick wslled tube, if the stresses a t the bore are
iequired with the greztest precision, primry cuts ere made at the outer surface
and secondary cuts a t the bore. Combinztioo of boring and turning is
necessry to reduce the error.
The precision of the method is determined by the accuracy of the
mexsurement of dimensional chmges and the care taken in the remod
of layers. Due to the stiffness of tb.e members, only small stri..in chznges
are produced for a given thickness of layer removed. Hence the thick-
ness of the removed layer 8r.d the stnin should be measured with tbe
gieetest precisior.. The length of the specimen. should be more than two
times its diameter. If not, end corrections ere necessary.
(iv) Hole drilling nzethod
This is a v;lluzble techn.ique because it can be used to survey the s w
*-
zt the surfzce of z large part. In. this method, ihe surface strains in the
nity of ir hole cre mezsured as the hole is drilled. This method Can b
ckssified as semi-destructive, since the hole produced may be repaid
,,,; structures. One of the techniques employed is to coat the swface
of the part vith brittle lacquer al?d drill sinall holes at various pomts of
interest. The crack pattern produced is used to estimate the direction,
,,pitude and nature of the biaxial resid~lalstresses.
For quantitative work stiaia gauge roscttes (3, 4 or G) are to be used,
ll!jless primary directions are hewn. In the latter case, strain gauges
locaredin appropriate directions are sufficient. 0n.e of the basic essump-
?ions made is th8.t thc ~csidnal stresses me con-stant in the region of the
drilled hole. The stress distribution. cam. be determined only for a limited
d ~ t hsince
, the surface strain chan.ges become negligible when the nmterial
is removed at a depth. below the surface of one to two h.ole diameters.
The precision of tb.e method is low, since the straim produced by
drilling are very small. There have been various attempts to improve the
sensitivity of the method and also to measure stresses as a function of
depth. The hole drilling method is not limitcd to parts with flat surfaccs
only. It can also be used in fillets and at other locatio~swhere section
changes are gradual.
II. Under the category of physical methods wc can list:
(i) X-ray diffraction method
(ii) mtra.sonic method and
(5) Magnetic anisotropy method.

(i) X-ray diffraction rnetlrod

This is a non-destructive method for thc measurement of residual
stresses. The principle of the method can. be briefly stated as follows.
The metallic materials which are crystalkine in nature difhact x-rays and
from the diffraction pattern interatomic spacings can be determined.
Stresses (either applied or residual) cause changes in the interplanar spachgs
fiereby affecting the diffrzction pattern. Shesses within the elzstic limit
cause a shift iv. the position of the diffraction peak. From this peak shift,
si-min (and hence stress) can be calculated.
The method can. be used either to fmd out the sum of principal stresses
at 2 point or the stress along any particular direction. .The direction and
W@tude of the principal stresses can also be determined.

Since the penetrating pocGer of x-rays used for diffraction work is small,
safely assume biaxial stress field. The diffraction inf~rmaeon
Ww from a layer of few e o n s thickness. Incidentally, this factor shows
328 C . BALASINGH
et al.
the importance of surface preparation. The accuracy with which stresses
can be determined depends on the accuracy with which small changes in
interplanar spacings can be measured. For this reason, high angle &f-
fraction lines are used. Lattfce faults and uniaxial plastic flow are found
to give spurious shifts. The f i s t one can be eIiminated by proper e*
mental procedure.
Either photo@aphic technique or diffractometer can be used. '&e
diEactometer is quite Wst and nowadays computer controlled stress diffractp
meters are being used in industries.
(ii) Ultrasonic method
Just like transparent mater:als are birefringent to a beam of polarised
light, stressed metals are birefringent to an ultrasonic (frequency of the
order of 5 MHz) sheer wave. A polarised shear wave passing througb surb
a material Is resolved into two componen.ts, which lie in the planes of the
primipal stresses. These two wave components travel at different velocitis,
which are dependent on the magnitude of the principal stresses. These
velocity changes are measured, and from that the stresses computed. How.
ever, complications atise due to the fact that birefringence is not only
ceused by stress, but is also produced s s a. result of anisotropy. There had
been many attempts to separate out the two effects and some are success-
ful. In futwe, this may prove to be a very powaful non-destructive methe
for residual stress analysis.

(iii) Magnetic anisotropy method

This is applicable only to ferromagnetic materials. The method in-
volves the measurement of the effective permeability over a range of frequency
of the applied alternating field. It is still in the development stage.

III. Chemical
Unde~this category, the main method is "stress corrosion method ".
Corrodents which cause cracking of the surface of certain metds w d
tensile stresses exist may be used to detect residual stresses. Though in
principle by controlling the conditions it is possible to get quantitativeda$,
it is highly unreliable in practice. Further, stresses below a particular vaf@
cannot be detected since no cracking will occur in that case. Since small
changes in composition of the metal or corrodent, temperature and xWd
anisotropy wr have large effect on the stress corrbsi~n behaviour, tfic
method is only qualitative,
 Residual Stresses in Iron Cnstings 329

Experiments on prototype ca~tings'-~"or thc evaluation of residual

and
)L,.~ss~ J the effect of different variables on their magnitudes will be too
expmsive. Hence the necessity for desiyning some simple structures.
The specimens used to investigate residual stresses can be classified
mro three groups.
J. Specimens designed in such a way ihat high castins stresses are
developed.
11. Blanks or approximate shapes (later machined) or exact sizes are cast,
and then subjected to various quencmg treatments and
111. Blanks or approximate shapes are cast and machined to final dimen-
dons. Preplanned stresses arc then introduced and the effect of
various relief treatments investigated.
Majority of the specimens comin-g under group I are designed on the basis
of differential cooling between parts of the casting. The grid castings were
developed on this basis. The shape of the specken used by Russell3 is as
shown in Fig. 1. In the thick mcmber, tensile and bending sti-esses (the
latter due to the weakness of the side members) are developed. A gap is
roilled in the thick member and the gap width is taken as a measure of
mess. The method is not very sensitive.

FTG.1 . Specimen used by Russel.

 Dodd" first tried to use a simple rectangular framework (Fig, 2).
~t was observed that the connecting members were subjected to high bending
stresses, and this in turn offered a partial relief of stress in the cenb;c

FIG. 2. Simple rectangular grid.

member. Following the discussion of Roth and Seumel,15 Dodd modifiut

the yokes, as shown in Rg. 3, to give the ends sufficient rigidity. In this
case the stress in the centre member approaches a state of uniaxial tension.
Parkins and Cowm15 and Kosowski1? used specimens of similar shape.
Kasch ard MikelosWs used a modiied rectangular framework desig
nated G-66-2B (Fig. 4). The desip looks a bit complicated, but it seem
to ensure unidi~ectiovalstress. The grid used by Patterson and Diet&'
is amlogous to the rectmgular framework. Here the thin outer memM
have a double trapezium-shaped cross-section whereas the centre me&
has a circular cross-section. The middle member is tapered in the
The yokes have a double T-shaped cross-section. Extensions p l d e d f n s P
the centra member can be used to clamp the specimen in tensile ,
machines. Amoxding to the authors the trapezium shape of the
 Residual Stresses in Iivn Castings

FIG. 3. Modified rectangular grid.

member prevents the rotation of the grid under the action of residual
stresses. The shape of the specimen suggested by Portcvin and Pomey2
is similar to the 0n.e used by Russell3 except tha.t the measuremen.ts mere
carried out on thinner members and that distance between two points before
and after cutting were measured instead of gav width. Triangular grid
castings were used by the Techwical S u b ~ o m m i t t e e ~(T.S.
0 ~ ~32)
~ appointed
by the Institute of British Foundrymen. A schematic drawing is shown
hFig. 5. The outer members and yoke, because of their high surface
e m to volume ratio, solidify fast and becom: rigid compared to the thick
centre member.
In the recta~gula~ as well as the triangular grids. strain gauges are
used to fuld out the stresses. One or two strain gauges are attached to
centre member and then it is sawed through. The strain indicated by
the w g e is straightaway used for the calculation of residual stress, if the
Wlue of Young's modulus is known. There are some diEculties in fixing
 Em.4. G - 6 2 B RecWUlar framework used by Kasch and M~klonis.

b 1-: "
,

. .
FP. 5. Trianplar Grid,
 Residual Stresses in e o n Castings 333
be co~,rectvalue of Young's modulus as will be evident from the discus-
,ions later. Kasch and Mikelonisl%suggest another method which could
be adopted by Foundries not in. a positio~?.to afford the strain gauge equip
mnt. Patterson and Dietzel,l9 in add.ition to determining the stresses
by the strain gauge techn.ique, also used a method called the deerence
,thod. However, the validity of the method is questionable.
The drawbacks of the grid castiqs are : (i) Th: microstructure of thick
thin members are usually not identical with the result that the value
Young's modulus may not be the same for different park. (ji) YoungYs
wldulusis not a constant for cast iron., but it depcn-dson the strain. Assump-
tion of a mean. value for Young's modulus may lead to errors. (iii) There
is always the likelihood of some bending of the members, which cannot be
taken into acc0un.t by employing the present methods of calculation of
stress.
The second drawback can be overcome by plotting a stress-strain
curve and reading off the stress corresponding to the strain observed.
%ere. is another procedure which doesn't involve the measurement of
strains. The centre member 1s sawn slowly until fracture occurs, If
the area of fracture is a, thc stresa a is calculated using the formula

where P and A are the tensile strength an.d the area of cross-section res-
pectively of the centre member.

Usually cylindrical rods were poured along with the grids. From
these rods specimens for tensile test and bending test were machined.
Average value of E (Young's modulus) c m be determined from eech test.
In addition to these, the value of E can be calculated from the graphite
content also. Patterson and Dietzel used these 3 E values for the calm-
lation of stress and tried to arrivc at a best fit which will gi3e smooth varia-
tion with other variables studled.
TWOother types of specimens can also be included in this group.
first is the double flanged barsz0 used by T.S. 32 to study the effect of
W ~ l dhindrance on the magnitude of residual stress. Next is the large
T %tion castings used by Girschovi& and S ~ ~ n o v s ktoi i investigate
~~
fhsting distorrion,
334 C . BALA~INGH
ef al.
The type of speciniens coming under group 11 are usually either
cylinders or hollow cylinder^.^ All the specimens were cast to approxi.
mate sizes, machined to final dimensions and subjected to an initial thermal
stress-relief treatment. Patteison and others used the cylindrical samples
to study the effect of various quenching operations on the magnitude of the
residual stresses, and also to investigate how the mechanical propertier
were altered by the residual stresses. The hollow cylinders, after heating
to a particular temperature were cooled differently and the effect of theje
on the magnitude and distribution of residual stresses studied.
Under group I11 we can list the ring specimens used by Hallett and
Wing,z3 Kotsyubinskii et and Zeppehuer and BrezinaZ5 and the
rectangular bars @ent transversely by a stinup using distance pieces at
ends) used by Benson and Allisonzs and Tattle.=' The ring specimens
were machined from hollow cylinders. A slit was made in the ring and a
wedge introduced so that known amount of stress can be introduced. In
this loaded condition the rings were subjected to various relief treatments.
From a measurement of the width of the slit before and after treatmmt
the percentage of stress relief can be easily calculated. Hallet and Wing13
used the following formula

where R is the percentage of stress relief ;

G,-width of the gap after removal from the spacer wedge;
GI-width of the original gap;
W-width of the spacer wedge;
P-plastic deformation at room temperature = (G, - GJ at RT
if no plastic deformation occurs at room temperature, P =O. Zeppelzauer
and Brezinaes neglect P. Both the groups of imestigators were study&?
the effect of different heat treatments on the magnitude of relief of stress
in iron castings of different compositions.
Some of the objections raised against this type of specimens are:
(i) l h e stress distribution across the thickness of the &-g va.g
continuously through the neutral axis.
(ii) The diffmence in coefkients of thermal expansion of the d d a i
gf tfie tin$ and of the wedge may introduce qdditiowl stresses,
 Residual Stresses in Iron Castings 335
(iii) Whereas actual castings are not restrained while heating, the
wedge, at least partially, puts a geometrical restraint.
Some investigators feel that it is wise lo draw a distir.ctiot? between the
lebxation of exteinally applied stress and the relaxatio~~ of inherent casting

KotsyubinsKii et aLZ4used cylindrical specimens also in addition to

dngs. The specimens were held in rigid holders an.d known stresses
introduced. Str'ess lelexetion over prolonged periods were studied.
Another type of ~ p e c i m e n ~ ~i ~ 9 , ~ Qin Fig. 6. This has been
s ~shown
mainly used to study the effect of vibration. on the relief of stress. The

FIG. 6. Cantilever type specimen.

Fcimea consists of a rigid central portion with 4 projecting cantilever

h n ~ The. cantilever beams form the w0rkir.g section of the specimen.
Re stresses are set up by inserting calibrated pins in the gaps. The pins
Were fixed to rigid supports. Vibrztory stress is applied in the form of
cantrated load 5 P in the rigid centre section. The vibratory load can.
be checked by monitoring the signals from the strain gauges attached to
the Wtilever zrms to an ~scilloscope. The gap widths are weasured
wore loading and also after fernoval of the pin subsequent to vibration
h w t . The change in gap width is used to calculate the relief of stress
as well as distortion.
The objection raised against the ring specimens can also be raised here.
Though this type of specimen has been used to study the effect of vibration
only, there is no reason why it should not be used to study stress relief by
other methods, for example thermal stress-relief.
The main advantage of specimens ur.der group 111 is that one can
introduce known stresses and then study how different treatments relime
the stresses. Also data collection is easier.

CAUSE^ AND EFFECT VARIABLES

OF DIFFERENT ON THE MAGNITUDE OF
STRESSESIN CASTINGS
RESIDUAL

A. Causes
The residual stresses in castings arise due to inhomogeneous plastic defor-
m~tions.While cooling to room teqperature, in order to enable the parts
of the casting to fit together elastic strains corresponding to the residual
stresses are introduced. Non-uniform deformations can be expected he
to the following reasons.
(i) Va~iationsin cooling in different parts of the casting.
(ii) Hindrance to free contraction by the mould material-often referred
to as mould restraint.
(iii) Phase transformations accompanied by volume changes.
Residual stresses can also be set up due to temperature gradients
existing from the surface to the centre of casting during cooling. Here
stresses are related to the overall rate of cooling of the casting. The sttessa
arising due to this cause attain appreciable magnitude, only under eon&
tions of rapid cooling (for example, quenching). Compositional and
structural heterogeneity can also cause stresses.
The relative contributions of the.above three factors to the final residual
stress depend on the shape and material of she casting, and other found0
variables. Phase transfbrmations and mould hindrance can cause, appre-
ciable amount of stress only if temperature difference exists bet- pa*
of the casting while cooling.

B. Effect of Different Variables on the Magnitude of Residual Stress

The factors which influence the magnitude of residual stress can be
classified into three categories :
(i) Composition, melt treatment and .pouring temperature.
 Residunl Stresses in Iron Castings

(ii) Mould materials afld propertics.

(iii) Other foundry variables such as hock-out.
Let us consider one by one.
(,) Composition, melt treatment and pouring temperature
In case of cast iron, we can consider the following compositional
variables :
(a) Carbon and silicon content ;
(b) Phosphorous content ;
(c) Sulphur content; .
(d} Manganese content.
(4) Carbon and silicon content
Not much literature is available regarding the effect of composition
and inoculation on the residual stresses in cast iron. Girshovich and
simanovskiiz%easured deflections in T section castings of grey cast iron
of different compositions. As the total carbon content was raised from
2.5 to 3.5% at a silicon content of 2 . I%, the deflection observed changed
from negative to a positive value. At a particular carbon content, zero
&kction was observed. Similarly at a constant carbon content, an
increase in silicon content from 1 . 7 to 2.7% made the deflec-
tion change from negative to positive. At a particular value of sificon
content the deflection becomes zero. As the carbon equivalent is increased,
the deflection red~~ces,passes through zero from negative to positive.
Further increase in the carbon and silicon contents increase the deflection.
Pattersow and Dietzcllg studied the effect of carbon and silicon con-
tents using rectangular grid castings. They poured six types of cast iron
viith different carbon and silicon contents. The results were discussed in
terms of K factor and the degree of saturation. The degree of saturation
Sc is calculated from the relation

and the K factor given by

33s C . BALASINGHet al.
the degree of saturation was increased, the residual strcsses decrmed,
For the same degree of saturation higher stress values were observed for
castings with hlgher K factor. Further they observed a linear relationship
between the tensde strength (or bendmg shength) and residual stresses,
prov~dedthe melt treatments were the same.

(6) Phosphorous content

In case of inoculated iron, the residual stresses increased with increase
in phosphorous content, if the E values were estimated from the ben&ng
tcst. On the other hmd, if E values determined from the graphite conteni
were used for thc celculatiolz of h'esidual stress, the residual stresses
increased slightly up to a P content of 0.49% and thereafter decreased
In case of uninoculated iron, tbe residual stresses increased if.E values esti-
mtted from giaphite content and tensile test were used. Experiments with
solid cylinders que~~chcd from high temperatures also showed some incruse
in residual stress with increase i~the phosphorous content.

(c) SuIPJIw content

To study the effect of sulphur content on ~esidualstress, irons inocu-
lated with calcium silicide were used by Patterson and Dietzel. The mi-
dual stress, cakulated with the E. velues obtained f r o p graphite content,
increased as the sulphur c o n t e ~ t was increased. The authors attribute
this to the carbide stabilizing effect of sulphur in cast iron. The residual
stress values, calculated with .E obtained from bending and tensile teds,
decreased slightly with increasing sulphur content, up to 0.1% and thereafter
increased.
(d) Manganese content
The iron used was not inoculated. The variation of residual stresj
with manganese content shows a maximum and a minimum. The reasons
for the minimum and maximum me thought to be as follows. Up to 0.5%
M3 the fraction of combined carbon decreases. Thjs muses, a decrew
in the strength of the material with the result that the rapidly cooling ibi@
members me r.ot in 2 position to deform plestically tbe thick member
Hence the stresses ere lower. Above 0.6% MII the fraction of combined
carbon increases, strength increases 2nd residual stress increases Rm
0.8% Mn onwards tramformation stresses become effective which in this
case reduces the finel value of residual stiess. This accounts for the 1118s
mum value observed.
 Residual Stresses in Iron Castings 339

p2,ttel.son and Dietzel ir.oculatcd the irons with fe~.rosiiiconand calcium

,i1icide, Inoculation with celcium silicide resulted in higher stresses.
However, with iv.oculation, the slrellgtil of thc iron elso increased. There-
fore, rp,rher t h m comparing the absolute val~iesof residual stress, it is more
rnli&tering if the i'atios of tensile strenglh to residual stress are com-
pwd. This ratio was found t o be very high for irons inoculated with
czlcium silicide compared to uninoculated irons. This meiins inocula-
iion (calcium silicide) is favourable from the residual stress point of view
&o. On the other hand, inoculation with ferrosilicon resulted in r? lower
ntio of tensile strength to residm.1 stress, i.e., the residual stress has increased,
hut the tende strength has not i ~ c r c a s e dto 1ha.t extent.

Po~~ringTemperature
Angus end Tortkl observe that low pouring temperalures give rise t o
h g 2nd blow hole d-efects. These defects czn provide stress raisers and
i~itiatecracks. According to them, relatively high p o u ~ % ~ temperature
g
will tend to reduce temperature differences due t o difference in sections
md slow down the overall cooling rate. Patterson end Dietzel also put
formrd the same ergument.
However, the results obtained 2.re just the I-everse. Dodd14 observed
P. slight increase in residual stress with incsease in pouring temperature
(A{ alloy). T.S. 32 in their work with A1 alloy and steels0 observed an
increase in residual stress with increase in pouring temperature. Girscho-
rich eral. observed increase in deflection of their r section. castings (coat
iron) with increase in pouring temperature.

liii Mould materials nnd properties

[a) hfould strength
Doddli from his investigations o n RR 59 Aluminium alloy in sap.d
-ids came to the conclusion that residual stress was independent of
fnoddhzrdness. Further he couldn't find any variation in residual stress
'MI mO~ldstrength (dry).
T.S. 32 using the iriz.ngular grid c x s t i ~ gtl.ied to find the vz.lii?tiot? of
rc&hl stress with mould strength in case of A1 eHoy m d steel. They
fmrd th8.t the variation of stress with the.strength of the mould didn't f@llow
Consistent pattern. However it is worth noting that in their expen-
I.1.S.c.-10
 merits with flanged bars, though they didn't observe any residual stress,
after removal from the mould, the cestirg in the herd rammed mould
contrscted' more than the casting in the medium rammed mould'. nir
immediately suggests .that provided suflicient difference 'in cooling raIe,
exist, the &ting in a mould yith high >trength can give rise to hi&,
residual stresses.
Parkins and Cowan16 conducted experiments .with cast iron,. brass
and Y alloy in moulds of diffirent strength. Though in each case wide
variation in residual stress with mould strength was observed, no meanink
ful relationship could be established. Experiments proved that high t e m p
rature strength is important and residual. stl'esses increased as the hi&
temperature strength increased. .

(b) Moistrve content in the mould

Dodd and T.S. 32 found that residual stresses inci'eased .linearly with
the water. content in the mould. They used A1 alloy and steel. Kasch and
MikelorG observed slightly higher value of stresses for casting. in the dry
mould than those cast in the green sand mould.
. .
(iii) Other foun&y variables such ks knock-out '
.

It is a practice in foundries to remove the castings f?om the monld while

it is still in the hot condition. Dodd (A1 alloy), T.S. 18 (grey cast iron)"
and T.S. 32 (A1 alloj and steel) observed an increase in residual stress WIL
stripping time, i.e., the castings knocked out early had a lower amount of
residual stress. The explanation put forward is that when the castings
are knocked out early, the temperature difference betweep different @
of the castings drastically comes down which results in a lower m a p
tude of i-esidual stress.
But Kasch and MikeIonisl%bsel.ved the reverse in case of ductile iron.
They observed a stress of 4800 psi when the casting was cooled to rwnt
temperature in the mould while the casting shaken out when the t W
rature of the light and. heavy sections were above 1340" C, a. stress af
6960 psi was observed.
The observations of Timofeev et IS worth mentioning here.
were investigating why the centl'ifugally cast pipes failed more often tl@
the turn table cast pipes (material: grey cast iron). From the c h a
content and microstructure point of view the centrifugally cast pipes wat
superior. On close examination they found the reason to be due to *
f
 Residual Stresses in Iron Castings 341
kn,,&-outof the centrifugally cast pipes at hign temperatures. According
them, knock-out at high temperalu1.e~ in.duces high tangential stresses
"hi& the, failure during testipg. Slower cooling reduces the
m,&tude of tmgential stresses.
n u s the effect of knock-out OP. the magn.itude of residual stress is
well established and requires further detailed investigations.
he Technical Subcommittee T.S. 3 2 tried to find out whether the resi-
duzl stresses varied when the dimension of the rumer was changed. They
an increase in residue1 stress wirh ir.crease in the dialnetel of the
runn.er (triangular grid castings). I t is cot clear why this should increase
Lt all.
According to the Tech~.icalSuboommittee T.S. 32, there 1s a definite
relationship between the structure of the casting an.d thsresidual stresses
de~eloped. Wher. the side members of the grid showed heavy chill struc-
ture, the casting crzcked in the mould itself d ~ i eto high stresses. This
cracking usually occurs in the web e t the wider-end of the grid at.the junc-
tion between the .side members. Acc0rdip.g to Ar-gus and Tonks, use of
deweners may provide rt directioral dendritic fornution .which in turn
increases the proner.ess to cracking. Also the castings ,were more brittle
in places where the graphite was i n the undercooled form. ,

The stresses arising due to differen-t c o o l i ~ grr-tes i r thick a d thin

sections, with special reference to grid castings, is broadly explained as
follows. When the molten metel is poured the thir. outer members start
caoiing faster t h m the thicker member, with the result that the thimer
sections shrink faster than the thicker. Thus the forces become operetive
first of ell produce tensile stresses in the thin.ner outer membeix and
compressive sti.esses in the thick middle member. The yokes serlre to
m m i t the forces and may be subjected to a certain. amount of herding.
Now, the mechar.ical strength is a temperature dependwt property, m d
thin outer members 2.s a result of rapid cooling, come k s t within. the
rage wherein they can @.ke up elastic str2.j.n.s. When. the thin members enter
this the thick member, is still in. the plastic rapge, wherein it canmt
"PWrt my elastic strair.. TP. other words the lighter ,section.s compress
Md &form plastiw.lly the thick centre member. As coo1in.g progresses,
the thin members, attdn tempex,atures wherein the rate of cooling and
eonWuently c0ntraction.s become smalle~,. Up to this point the oukr
membe1.s are under tensile sti-ess. After a certain time, the cooli~~g 12.ia
of the thick and thin mcmbcrs bccome equal and fin.elly the rate of cooline
of the cenrral member exceeds that of the thinner members. Wher? the
centre member has cooled to s? tempei-ature wherein it c8.n take elatric
strains, a process of stress yeversa1 takes place. Some refer to it as themi
stress leversal. Filst the remile stresses in the outer Inembers decrease 7.~6
then become compressive. The thick portion tries to contract, but it is
restrained by the rigid outer frame. Thc effort by the centre member to
contract and thereby to pull to get he^, the stiff outer membecs is Yesirted
and this leads to temile stresses in the thick centre member. If the tensile
stress thus produced exceeds the tensile sti-ength at those temperatures.
cracking can occur. Or if the compressive stresses are high ir the thin
outer members, bucklirg of these parts cm occur. Or the czsting my
be left in a state of balarced stress system.

As the Subcommittee T.S. 32 puts it, the overall view is compl~pllwted

by the fact thzt the mechanical properties of ihe metal do not change
suddenly in the manner suggested. The change is giadual and sprezd
over a temperature range.
In the mechanism desci-ibed above, local plastic deformation of one
part relative to mother part of the casting is m i essentiel prerequisite. Let
us consider a case wherein plastic deformation doesr't occur, but all tbe
strains remained elastic from the moment of solidifieztion to cooling down
to room temperature. In tMs case the thinner sections, cooling rapidly,
develop tensile stress whereas the tNck portion, cooling slowly, develop
compressive stresses. But the strain p~oduced is now elestic. Later
when the thicker section cools faskr, the compressive st~.essesin the thicker
sections znd t5e tensile stresses in the thin sections decrease gradually. At
ell stzges the deformztiom are in the elastic range and completely balancd.
When the e a t i n g has cooled down to t o o n temperature, no residual stresses
would be pr-,se~.t.
If the mechanism of diffetential cooling (with local plastic &for-
mations') is operative, it is obvious that saesses will increase if the temp*
r m r e difference between thick and thin members are increased. This @
be brought out by casting conditions or increasing the diffeience in section
thicknesses. However the incresse in stress cannot go on indefinitell.
Wnen a certain value of cross-sectional ~ a t i o is crossed, though
stresses are produced in the thin outer members, this cannot cause mud
phstic deformation in the centte member, owing to its large area of
 Residual Stresses in Iron Cnsti~gs 343

,tion. This results in a lower residual stress in the centre member corn-
pared to that with a lowel ratio of Cross-sectional area. This is clearly
&monrtrated by the experiments of Dodd14 in case of At alloy castings in
mould (Fig. 7). The trend of result should equally hold good for
orher materials, though numerical values may be diffe~ent.

RATIO OF CROSS SECTION, ~ / A D

Fw.7. Relationship between stress in centre member and ratio of cross section of centre
and outer members.

T.S. 32 in their attempt to establish that the reverse is also true,

&rased the thickness of the centre member from veiy low values. As
expected, when the cross-sectional area of the centre member was lower
rhan that of the outer members, compressive residual stress was observed
in the centre member. As the thickness was increased, the compressive
stress decreased and finally became tensile, after passing through zero.
In materials which undergo phase changes in the solid state, accom-
wed by volume changes, transformation stresses also arise along with
[he rhermal stresses. The magnitude of stresses remaining at room tempe-
depends on whether these phase changes, with the associated volume
"ban@% occur at higher temperatures wherein the stresses are continuously
reduced by plastic deformation or whether the phase changes teke place
at s temperature when the m-terir.1 C ~ P . take up higher elastic strains.
Patterson and Dietzell3 point out, 2 . q alloying element which influences
the rate of trmsition of the elastic plastic beb.?.viour during deform~.ti~~,
alters the residual stresses also.
Pakins end Cowm16 conducted a series of elegmt experiments to under.
stand the mechznism of residuzl stress fo1-metion ill. s m d castings. The rec.
tzngul?~fremework castings were heated m d then cooledunder conditions
which would hzve been existing in e s m d mould. Differential cooling waj
obtained by winding the members with suitably thick esbestos 1,ope.
The temperatures of centre member and outer members were monitored
while cooling down from different temperatures. They were zble to
estzblish a lil:.ea relationship between the final residual stress w d rhe
mzximum temperature difference. For non-ferrous dloys it w2.s a mono-
tonic increase, whereas for c2st iron at a temperature difference of zbout
18O0C, a sharp increase wi?s observed (See Fig. 8). This discontinuity
is associated with the As1 trmsfo~mr?tion. The experiment elso brou&t
out the f a x that trmsform.tion taking plzce in one portion wzs able to
pli?stic?.lly deform another pzrt. When they cooled the sanples from
730°C or below, the transformtion in all the three members took place
mder cooling conditions which are similzr, and corsequently had no effect

2
z
'S
in
2
f2
I
%
W
E
"?

--. .. .. . TEMPERATURE DIFFERENCE. "C

Residual stressinlnx temperature diffelence relationship. Y-alloy framework

 Residual Stresses in Iron Castiizgs

.TEHPEIl*IURE DIFFERENCEc:.

FIG.8 b. Residual stress/max i e ~ p e r a t u r edifference relationship in cast iron frameworks

on the final vzlue of residual stress. However, while cooling from higher
temperatures, the deformations associated with the higher temperatures are
not completely reversed, which adds to the final value of stress.
Now coming to mould hindrance, as in the case of phase transfor-
mations, the final values of residual stresses are altered only if sufficient
temperature difference existed while cooling so that one part plastically
deformed compared to the other. Parkins and Cowan16 did a series of
experiments with rectangular grid castings of different materials using sands
of different strength. They summarise the results as follows : In non-
ferrous alloy frameworks, the stresses may be entirely attributed to the
temperature differences developed while cooling. In grey iron castings
the contributions due to phase transformations and sand hindrance may
be of the same order as that due to temperature differences.
T.S. 32 did experiments on cylindrical samples with heavy flanges at
each end, cast in horizontal position. After removing from the mould,
change in length occurred, but no residual stress could be observed. This
confirms the earlier arguements that elastically balanced strains should
disappear.

DO we want to remove the residual stresses completely? The answer

is we Would like to remove the dangerous residual stresses. One should
be clear as to the requirement. I s it just the removal of locked-in stresses
that is desired or is it the distortion which one wants to avoid. From earlier
 times it has been a practice to store the castings oul of doors for "wes.thei.
ing ", by which it was thought that the casting acquires dimensionel stabi
lity. Even now thc common recommendation one comes across is rough
machine, age, and finish machine.
If the d m is just reduction of residual stmses, therma.1 mnealing
presents itself as a proven technique. But considel'ip-g tbe zbility of the
casting to resist further distortion onc has to consider the stress rehxa.
tion resistance also. According to Novichkov3"t is necessa1.y to consider
the potential energy levels i ~ sd.dition
. Lo stress level and stress relaxation
resistance.
The two treditionrrl methods of redwing residual stresses are:
(i) Seasoning or weathering (natural ageing),

(ii) Stress-selief annealing.

Three more techniques me a k o beilig employed and/or investigated
nowadays. They are:
(iii) Thermal shock,
(iv) Vibration and
(v) Static overloading.

(i) Natural ageing

There is controversy as to the magnitude of stress relieved by this
process. However, everybody agrees that the mag~itudeis sma.11, but what
is important, a certain dimensional stability is obtained. Stress-relaxation
resistmce is increased a bit. According to Russe131, weathering for four
months relieves stress by about 15%. The experiments by TottlePiznd
Nallet% proved that weathering didn't relieve any stress, which ageing couldnl
achieve. Hallett and Wing2"n their experiments with different types of
iron were able to observe a m4ximum of 10% stress-relief by prolor-ged ageing.
(ii) Stress-relief annealing
This method hiis been in practice from very eai.1~d2.y~. The tcrnperarufs
to which particular materials ?.~eto be heeted can be found from the Md-
books. Still it is worthwhile to recall some of the sdient features brow&
forward by research workers. As the temperature is increased, no doubt,
the stresses w e relieved in a short dur?.tion, but at the same time strength
, 2nd hardness come down. This is undesirable. m e should use t e w
i:
S
 Residunl Stresses in Iron Castings 347
which reduce the stress considerably, but does not bring down the
mechanicalproperties. In case of iron castings there is one more fkctor to
he t;lken into account. Tf the temperature reeches the lower limit of graphi-
tistion rage, the accompanying exprmion may create new stress. The
ttmpep.ture chosen must be higher t h m when the plestic deformation23
be@ns. This temperature for grey iron is between 350 and 450" C .
&ny authors think that ?.n~.eelingabove 550" C mzy result in s change
of stmcture and deterioretion of mechanicel properties. Russel experi-
menting with various types of iron, both inoculated and uninoculeted,
observed no relief of stress up to 400' C, temperr?tures in excess of 525" C
were required to relieve 50% or more of 1.esidua1 stress, evep. a t 600' C the
residud stress was not completely relieved. At 600" C he observed s fall-
ing-off of Transverse rupture strength. His recommendetions are : rough
mxhine-anneal et temperatures between 550 and 600" C-dlow as much
time as possible before fini~h mechining.
The rate of heeting end coolirg zre importat?t. If high, especially in
complicated shapes, there is a risk of failu~e. Experimer.ts showed that
it is not actuzlly the rete of heating or cooling that is important, but the
uniformity of heating or cooling. Zeppelzmer and BrezinaZ5 recommend
the r?,te of cooling not to exceed 30 to 40°/hr. Simple shapes should be
cooled in furnace up to 300" C and complicetcd shapes up to 100°C.
K i ~ hsnd Mikelor.islR estimeted the stress I-elief obt~.i.inedby employ-
ing various heating cycles: (a) very fast heating rate, slow and fast cooling
a t e , (6) slow ;md f a t heeting retes, slow 2nd fast cooling ~ates. In (a)
they =me to the conclusion that 2. cooling rate of 100" F/hr was optimum.
The disastrous consequences of non-uniformity in heding rete was clearly
ihom by their experiments on grid czsti~gs. When the heating rate was
hk'hly non-uniform the grids cracked.
Mo~ingincrer.scs the temperature requiied for stress-relief. Some
%&ate this with the carbide stzbilizing influence of thc alloying ele-
ments. pol example Cr, Mo, Ni a ~ . dV shift the beginning of plastic dcfor-
wion to higher temperetures. Since Cr and Mo retard the decomposition
.ufementite at high temperetures, i t is possible to remove residual stresses
cast irons with these elements at high temperatures without sffecting
inrallo~
mechm.kal properties adversely.
temperatures recommended by different investigz-tors are slightly
H a k t and Wing obt~.ined 82% and 95% relief when annealing
Llsc-11
S T A ~cast iron r.t 600 m d 650" C respectively. 'For centrifugal cast iron
82% at 600" C and 97% at 650" C. For 33% Cr Cast iron the relief obtained
at 600° C was between 10 and 25%. At 650" C they obtained a stress lelief
of 90%. FO; Ni-resist at 600°C relief of 35% and a t 650°C 83% wee
observed. In case of 33% Cr cast iron, anneeling increased the tensile
strength slightly. But the tensile strength of Nibresist was found to
decrease by about 10%.
Zeppelzauer and BrezineZ5feel thit the clzssification of gey cast iron
into unalloyed, low-alloyed and high-alloyed, for the purpose of them1
stress-relief treatment is inadequate. On the basis of their investigations,
they classify a ~ ~ecommend
d temperatures as given in Table I. The
results given in a handbook are also included for comparison.

Sl. Chssificztion Temp. required for Data given in

No. stress reduction of VDG leaflet

1. Unalloyed 550 590 500 to 550

2. Low alloyed with

Cr or Ni or Sn

3. Low alloy with Cu or Mo 590 615 )

4. High alloy with Cr or Ni 595 625 600 to 650

S.G. iron hzs a higher residual stress. Transition occurs in the

nture range 5a@-6aOo C. According to Zingg,l2 a t a temperature d
550" C considerable stress reduction takes place. Some books rm*
mend a temperature of 68OoC. But a t this temperature bardness and
tensile strength are reduced. For ferritic S.G. iron the temperature mBt
should be between 585 and 615" C, whereas for pearlitic S.G. iron, be-
570 a d 595" C. In case of ferritic iron, annealing results in a lo&gd
hardness UTS values, but increase in Young's modulus and elon&attan.
This is attributed to the dissolution of the small fraction of pearlite in tht
 Residual Stresses in Iron Castings 349

In case of pearlitic S G iron, E ircreases slightly, harhess e.nd

UTS values are altered very negligibly. No change in structure was
observed.
Residual stresses resultirg from compositional and structural hetero-
geneity in castings cannot be removed by thermal annealing."
(iii) Thermal S ~ Z O C ~
Thermal shock method consists in either rapid heating or lapid cooling
or both. In natural ageing, stress is relieved to a sm'll degree, and stress
&xation resistance increased slightly. Regular high temperature a~neal-
ing brings down the level of residual stress drastically, but unfortunately
thestress relaxation resistance also comes down. Subseq~1en.tmachining may
cause renewed distortion in both the cases. In this point the thermal shock
method scores over the other two. Thermal shock raises the stress relaxa-
tion resistance considerably so that subsequent machining may not cause
renewed distortion.
Kosowski17 puts forward two more points in. favour of thermal shock.
In regular thermal annealing pzrtial decomposition of the cementite can
take place, which lowers the mechanical properties. Another point is the
lime factor.
Rapid heating ceuses permanent elongation of the casting with thicker
sections and contractions of those elements with thinner sections. Rapid
cooling causes deformation in the opposite direction. The rapid cooling
process reduces the time necessary for repeating the process. Kosowski
heated the castings in a muffled furnace to temperatures of 500, 600, 700
md 800" C respectively. After a holding time of about 5 to 8 mts the
castings were taken out from the furnace and immersed in water at 20°C.
This cycle wcs repeated till appreciable reduction in stress was obtained.
As the tempnature of the furnace (and hence that of the casting before
quenching in water) was mcreased, the number of cycles required to rebeve
the stresses decreased. It w2.s found that a t a furnace temperature of 500" C
(orsting temperature 300" C) even after 50 cyoles, appreciable amount of
stress remained, whereas at a furllace temperature of 880" C (CT 700°C)
&~mtcomplete relief of stress was observed after 3 cycles. Theseexperi-
mnts were done with rectzngular grid castings. According to him for
shape of the casting, ihe heating and cooling rates, and the number
Of cycles should be established separately.

Gechikov and Kot~yubinskii3~

had done fairly detailed analysis regard-
& the stress-relief by the thermal shock method. They established that
 the thernxl shock method cap be applied to 2. feirly wide rmge of castings,
They bring in the overload fector K which turns out to be one of the most
important parameters controlling the dimensionel sti:bilisation of castingr
by the thermal shock method. K is the ratio of the stress induced by
thermal shock to the initial stress. A linear relationship was obtained
between the reduction in distortion and the K f x t o r . At a K value of
1.5 to 2.0 the subsequent stress relaxetion was reduced to 1/10th its original
mggnitude. Surprisingly, the tempereture he employed wzs between2m
and 400°C only, whereas Kosowski used higher temperatures.
To get the correct K factor, it is necessaly to select the correct furnze
temperature and holding time. The zuthors give a set of nomograms
so thet the required conditions can be selected w ~ t hease. They h e
indiceted how to proceed methodically.
The method can be applied safely to czs~ings contci~lirgresidual
stress not exceeding 25% of the tensile strengrh of the iron in bending
If it exceeds this value, there is a likelihood of casting failure during the
thermal shock treetment. In those cases they recommend a low tempe-
rature annealing to bring down the residual stress levcl to less than 25:,
tensile strength, to precede the shock treztmerr. The authors subjecteda
large machine casting to this treatment and didn't observe any distortion.
The nomograms, etc. developed are useful for small 2nd medium casting
weighing up to 3 to 4 tomes. For large castings further considerationi
are necessary.
(iv) Stress-relief by vibration
The advantages of using the mechanical vibration for the relief
stress are :
(a) Economy : The cost of vibretory equipment is low compared ta
furnaces, especially when the parts are of large dimension. Worzng
is also less for vibrato~yequipment compared to the equipment for thermal
stress relief.
(b) T h e .factor : The stress relief by vibratioll is accomplished in a
short period compared to thermal stiess relief.
(c) It occupies less floor space and is easily portable.
(d)NO oxide scale formation on the components strewrelieved
(e) No reduction in strength and hardness values.
Glancing through the above-mentioned advantages, one m y be
tempted to exclaim, go in for vibration only, why resort to the old t h d
5
 Residual Stresses in Iron Castings 351
at all ! But the research on this particular method hes not
rezched a stage that one can pick up a component, subject to vibra-
;ions as prescribed and then expect for 100% stress-relief. Conflicting
opinionsare expressed in the literature-one group saying that the stress-
reliefis complete, while the other group contend that they could not detect
relief of stress at all. A parallel to this call be found ir the czse of
relief of residual stresses in weldings by the vibration technique. Here also
Same controversy exists.
One of the earliest investig?.tions was by McGoldrick and Saunders.3'
They were not able to establish the zmount of stress-~elief obrained by
vibration, however, a significant increase in dimensional stzbility was ob-
served. The third reportsS of the Subcommittee T.S. 32 was entirely devoted
to vib~etorystress-relief. They took triangular grid castings (grey cast
iron) ~ n subjected
d them to v i b ~ ~ t i oby
n three different methods (i) Pneumatic
vibrztor-3587 cycles/min, (ii) Vibrator used for knock-out grid-lo00
cyclesjmin, (iii) Barrelling-32 rev/min. In F.O czse did they observe any
stress relief. In another series they used loaded jigs instead of grids. Here
dso they didn't observe any relief of stress. They concluded that no useful
relief of stress could be achieved by the simple vibration treatments applied.
However, the scatter in the numerical values is too high, sa that one.has to
?&e the conclusions arrived a t with caution. Also the mode of mount-
ing tbe grid is not indicated. This is very important as will be shown later.
Gut mounted triangular grid castings of grey iron with about 2.5 tons/
in.% tensile residual stress in the centrrl member on an Amsler pulsator.
Initizlly a compressive load of about 5 tons was applied. A further fluctuat-
~ superposed a t a frequency of 151CjS. He
ing stress of 1-4 t ~ n r / i n .was
determined the residual stress-relief as a function of frequency, number of
vibrations, fluctuating load and initial compressive stress. In no czse did
he observe any decrease in residual stress. On the contrary he observed
En innease in stress as a function of the fluctuating load.
Kasch and Mikelonis18 subjected G-66-2B test castings of iron to sonic
vibration. The castings were rigidly clamped to a steel plate .to which the
vihm Ws?s attached. The frequency used was 160 CIS. They observed
a shs-relief of the order of 6 t o lo%, compared to 74% stress reduction
if the castings were treated a t 1100' F. They also made bigger sizecastings
Wd subjected them to resonant vibration. There too they didn't observe
eppreciable reduction of stress. However, from theu description of
c w g , it seems that the casting was simply executing leciprocatory
IBDtiO% Without behg subject to any flexural stress. In that case, one
352 C. RALASINGH et al.
cannot expect any relief of stress. Buehler and T o e n s h ~ f fcame
~ ~ to ~e
conclusion that vibration is incapable of removing residual stresses.
The approach by Kotsyubinskii et al.," Adoyan et al.Pg,30 and L ~ P
shin40 seems to be more pragmatic. Citing the controversy existing in tl.,?
literature thcy conducted some experiments to clarify thc situation. ne).
djdx't use specimens which contained stresses in the as-cast conditior.
But stresses were introduced externally. The shape of the specimens used
by Kotspbinskii and Adoym is shown in Fig. 6. The distance between
the cantilever arms were mezsured accurately, rtnd change in this disbnancr
is used to compute the relief of stress.
Kotsyubinskii et a1.28 did experiments on Sch 21-40 cast iron. niev
found thzt the relief of stress was strongly dependent on the amplituds of
vibration; higher the amplitude, higher the stress-relief obtained. How.
ever, one cannot go on increasing the amplitude sjnce fatigue failure m.
start %,hen the limit is messed. The bulk of the stress reduction takes phcr
during the first hour; extending the time leads to further reduction, but
the amount of stress relieved is relatively small. It is better to limit tt:
time to 1 or I+ hours when 90% of stress reduction correspondi~gto thc
particular experimental conditions would have occurred. Th:s obser-
vation is validated by other investigators also, though some vibrate it as
long as 3 hrs. The manufacturers of the vibratory stress-relief equipmem
recommend time of only 1 hr or below.
When the frequency was wried from 7,600 to 12,000 c/min at approxi-
mately constant amplitude, the relief remained same indicating that at least
in this range the effect is independent of frequency. It is dependent only
on amplitude m d time. The maximum stress relaxation obtained wac
only about 25%. But the prime advantage is that, after vibration treat-
ment the component becomes stabilised. After the vibration treatment,
there is n o tendency to redistribute the residual stress, whereas in speci-
mens not subjected to vibration treatment, the stress relaxztion conti-
nues for long pe~iods.
Adoyan et al.29,30also conducted experiments on same type of s&-
mens. Then mzin aim was to establish the conditions to stress relieve spai-
mens with different initial stresses. They d ~ dnot report how much s b a
was relteved. Paramount importance was given to the distortion occurrine
after vlbrr?.t~ontreatment at different condltlons. The distortion m m w -
ments were continued up to 3 or 4 months.
In this context they introduce a parameter K called the overload fztof,
The overload factor is defined as the ratio of the vibration amplitude to
 Residual Stresses in Iron Castings 353

Initial stress. The specimens were give11 different initilll stresses, 6, 8, 10

l j kg/mm2. The vibration i?mplitude was i 3 kg/mm3 so thzt the
percentageoverload factor varied from 5 to 20 (Kfrom 0.5 to 0.2). Vibra-
I,,, time was 3 hr at a frequency of 50 CIS. The plot of distortion vs
,me for the spechens treated this way is shown in Fig. 9. As the percen-

TIME . DAY5 .-
Em. 9. Distortion of vibrated specimens.
~p overload factor increased from 20 to 50 (factor of 2.5) the subsequent
distortion reduced by a factor of 10. In their second investigation also
they found identical resulti. The surprising result is tht in spite of differ-
ences m the initial stress values and vibrational amplitudes, the distortion
curves correspond strictly to the K value. Whenever K L 0.45 there is
practically no distortion. Hence to fully stabilise the castings correct
choice of K is essential.
If the initial stress value is known, immediately one can fix the vibra-
tion amplitude to get the correct K value, subject to the condition that the
farigue limit is not crossed. For this the authors classify the castings into
two groups: (a) the case where the residual stresses are below 0.25 times
the tensile strength. Here vibration treatment with appropriate overload
c?.n be straightaway applied. (b) The case where the residual stress exceeds
0.25 times the tensile strength of the iron. Two alternative procedures are
recommended in such cases: (i) Apply the vibration amplitude above the
fatiwe limit for a short duration. The time should be low. This procedure
W'be necessary especially in castings with a high level of residual stress.
@)Where the residual stresses are in the range 0.25-0.75 the tensile strength,
first subject the castings to thermal annealing treatment so that the resi-
dual stress is brought to the 0.1 to 0.25 tensile strength range. Then
subkt them to vibrational t~eatmentwith appropriate K factor.
 Accordirg to Lokshin40 useful effects of vibration are to be expected
only when the object undergoes strain with every cycle of vibration. ~h~
component should be clzmped in such a way that it is free to deform under
elastic vibrations. The points of support should be as localised as possible.
If the component performs simply recip~ocatory motion along with the
vibrator, no strcss relaxation could be expected. Alloys with lo^ relamtion
criteria will respond more readily to vibration treatment. Lokshin's eqerj-
ments were mainly with aluminium alloy specimens. He observed sipi-
ficant reduction in residual stress (up to 74%) of components vibrated
under resonant conditions. According to him vibration treatment under
sub-resonznt conditions is not that effective.

Now let us consider a case where the vibration was applied to aproto-
type casting. The pzrt chosen by Skaznenmikal was the top trble of a
circular grinder of Sch 21-40 iron. The rough casting weighed 175kg.
which came down to 118-5kg after machining. The slideways were
machined 2v.d finish ground prior to vibretion ageing, leaving 2n inllow
znce of 0 - 3 mm for finishing. After vibrating under 3 different condi-
tions, the cestings were set up on special racks for impection and measure-
ment of distortion. Simultar.eously observations werc also carried out on
specimens (a) which were not given zny treatment a t all and (b) which were
subjected to thermal ageing. The distortion in untreated samples (natural
ageing) continued for 18 months or more; in thermally aged samples it
continued for a period of 6-7 months whereas in samples subjected to vibrz-
tion treatment, the distortion stopped after 33 months. Thus for &is
particular component, u~lderthe above-mentioned conditions of vibratioa
treatnient, the time required for the cessation of distortion was half that
required alter thermal ageing and a quarter of ihat required for nzturd
ageing.

The rcasoc for the contimance of distortion for a n appreciable the

after vibration may be either due to insufficient application of the overhi
factor or the method of clzmping or both. insufficient K will always
a problem, so long as one does not know what the initial stress level is.
This immediately brings forward one of the drawbacks of the m e w .
At least approximate knowledge of the residual stress is necessary ~O~PPLT
the correct amplitude. In those cases it is of immense help if tile m2@-
tude of residual stresses czn. be predicted from the geometrical shape of
the casting and the cooling conditions. A fe* attempts have been
in this directionPP9 45
 Residual Stresses in Iron Castings 355

IF. this method the compon-ent is loadcd in the static conditiol1.4~ The
directionof the applied load (called overloed) sh.ould be i i ~ the same direc-
tion 8s the residual stress. Kotsyubinskii e,t alas compz~ing the static
overloadingand vibration techniques, found out thzt for the same over-
b;ud, h e percentage of stress reduced by vibr~tionis higher than thet
by static overloading. However, the seme degree of stress-relief
be ob~inedby increesing the m a g ~ i t u d eof edditional imposed static
Jwd. The advantage of this method over th.e vibration techl~.iq~e is that
one need not be haunted by tb.e fear of crossii~.gthe fztigue limit. T i is only
nrcessary thet the RS -I-static load should be with.kJ. the stetic skength
of the materials, which is high.
Static overloading, like in the casc of vibrational ageing, is work
hrdened so that stress relaxation uesjstarl.ce 1s increased.29 In a casting
ofcomplex shape, it is rether v e q difficult to estimate the sti-essesin different
sectioos and thcn apply static loading of the same sign as the residual stress
at different sections. The method doesn't seem to be popular.
N o ~ i c h k o v ~summarises
~ what different treatme17ts achieve. The
parameters considered are residual stress level, stress relaxation resistance
and potential energy.
Process Residual Stress relax- Potentla1 Remarks
stress ation resistance energy

Yatural ageing Lowers RS Apprcciably (?) Lowered

by 7-20% raised slightly
Annealing in Lowered by Decl-eased Lowered
the range 85-90 % appreciably
54)-MX)"c

Hearing in the Lowered Apprcciably Lowered Practical

presence of raised difficulties
elartic vibra-
tion
Thermal shock Lowered Increased Raised
SMic oover load Lowered but Increased Difficulty in
and vibration amount case of com-
depends on plex castmgs
different para-
meters

L L L.-12
 356 C. BALASIN
GH et ai.

Heatkg ir the piesence of vibration has 811 the good qualities. Thou*
beset with prsctical difficulties it may be worthwhile to explore &is fie!$
further. Novichkov recommends another procedui'e ; ihe~'malcyc~p.~ in
the temperature range 200 to 280° C. According to him this is mwh
more effwtive in giving dimensioml stabilisation to the castings th;m
ing at high temperatutes.
Adoyan et ~ 1 also
. ~recommend
~ low temperature 220-300" C to get
better dimensionzl stability. They say, hect txatment below 3mec
produces substantid stren.gthening. Residual str:ss is lowered by about
8% orrly.
Summing up, we see th2.t ciifferent methods have different adVaF,.
t a p . Though beset with precticzl d.ifEculties, combination of two or
more methods seems to be more effective. Though the mechanismof
stress-relief by various met.ho0s is not clearly understood, it is legitimte
to expect the graphite particles to play some role in the relief of residud
st1,esses. The influerxe of gi-aphite particles on tbe plastic defo?mti~
of iron castings has been analysed in
Some practical hints to produce stress-free castings
Rather than producing castings with high stresses and then subjecw
them to stre~sr~elieftreatment, the ideal thing will be to produce castins
without any or very low residual stress. The detailed analysis of causes
and the mechanism of residual foi'mation are helpful in this regard. Alsc
a ffiorough kit-owledge of how the residual stresses are altered by diffaeot
parameters will help to choose conditions wherein residual stresses are
minimum.
It has already been established that temperature diffeie~ceexistiq
in the casting while cooling is the main cause for the fo~mztionof reslduq
stless. b - y step taken during casting t o reduce this temperature differam
will result in a reduction of stress. One of the techniques usually adopted
is casting into hot moulds. T.S. 32 has found substan.tla1 reduction bj'
this procedure. When (grey cast iron) cast into moulds at room t e w
rature, 200" C and 400"C residual stresses of 8 - 90 ton~lin.~, 5.18 tonsrm.'
and 3.59 tons/in.~espectively were observed. Knock-out is already
cussed. Other procedures are: using sand of 5igh thermal conductiktf
, or metallic chills around heavy portions, running the casting through
thinner portions, etc. Some of these steps m y create other fou&
problems. So one has to be choosy. Kotsyubinskiirn has devdoped a
 Residual Stresses in Iron castings 357
,!$tern which will measure temperature at different parts of
,hped castings 2nd apply forced cooling wherever necessary.

The =uses of residual stress formation seems to be fairly well uader-

s*ood, Due to a variety of reasons it m y not be possible to take all neces-
steps to eliminate residual stresses durir.g the casting stage itself.
~ h u swe ate left with cesti1l.g~co~.tzi~.ing
different amounts of stress, both
in mpitude and direction, which varies with.iv. a ccastir.g and from c2stiF.g
,a~tjng. It becomes necessary to relieve these stresses. Though 2. number
of metl?ods we being used, there ere some limitations to each method.
The mechanism behind stress-relief is not understood very clezrly. This
requires further study.

I , A n w , H. T. and Proc. Inst. Brit. Foundrymen, 1957, 50, 14-18

Tonks, W. G.
2. Portevin, A.andPomey, .I.Proc. Inst. Brit. Found~ymen, 1955, 48, A60-72.

3. Patterson, W. and Giesserei (Tech-wise-Beiheffe), 1966, 18 (3), 151-163

Dktzel, G.
4. SAE Information Report, J 936, 1965.
5. SAE Information Reporr, TR-414, 1960.
6. Erickion, M. A. Symposium on Testing of Cast Iron with SR-4 Type of Goge,
pp 2-10.
ASTM special Technical Publication No. 97 (1949).

Symposium on Testing of Cast Iron wifh SR-4 Type of Gage,

pp 11-15.
ASTM Speci?,l Technical Publication No. 97 1949.
I. Stocken, S. J. and Symposium on Testing of Cast Iron wifh SR-4 Type of Gage,
Lomie, H. W. PP. 16-18.
ASTM Special Technical Public?.tion No. 97, 1949.
Symposium on Testing of Cast Iron with SR-4 5 p e of G a ~ e ,
pp. 68-69.
ASTM Special Technical Pr.blicttion No. 97, 1949.
Symposixm on Testing of Clst Iron with SR-4type of Gage,
pp. 73-83.
ASTM Special Technical Ptrblication No. 97, 1949.
Proc, Inst. Brit. Foundrymen, 1935-36, 29, 650-684,

12. Zings, E.
13. Russel, P. A.
14. Dodd, R. A.
15. Roth 2nd Setmel
16. Parkins, R. N.and
C o w ~ nA.
,
17. Kosowski, A.
18. Kasch, F. E. and
Mikelonis, P. J.
19. P~tterson,W. and
Dietzel, G .
20. Firs1 Report of
Subconunittee T.S. 32
21. Second Report of
Snbcommittee T.S. 32
22. Girshovich, N. G. and
Simanovskii, M. P.
23, Halktt, M. M. and
Wiaz, P. D.
24. Kotsy.~binskii,0. YE.,
Obernun, I. Y a . and
Grechikov, A. M.
25. Zeppelzluer, K. and
Brezin~,P.
26. Benwn, L. E. and
Allison, H.
27. Tattle, C. K.
28. Kotsy,?binskii, 0. Yu.
et 61.
29. Adoyaq G . A. et al.
30. Adoyan, G. A. et at.
31. Rewri of Subwmmittee
T.S. 18
32. Timofeev, A. A. et al.
33. Novichkov, p. V,
k
C. B A L A ~ I N G H et ai.
Sulzer Tech. Rev., 1962, 44(4), 25-34.
Proc. Inst. Brit. kbnndrymen, 194546, 39, ~ 1 8 5 ,
J. Inst. Metals, 1952-53, 81, 77-81.
Giesserei, 1943, 30, 153.
J. Inst. Metals, 1953-54, 82, 1-8.
Foundry Trade Journal, 1968, 125, 957-958,
Modem Cartings, 1969, 55, 77-89.
Giesserei (Tech-wiss-Beihefie). 1966, 18 (3); 165-178,
Foundry Trade Journal, 1952, 93, 4 7 1 4 7 , 505-509.
Foundry Trade Journal, 1956, 101, 19-27,
Russ. Cast. Prod., 1963, pp. 72-76.
Prac. Inst. Brit. Foundrymen, 1949, 42, M8-3s.
Russ. Cast. Prod., 1962, p. 190.
Giesserei-Praxis, 1966, 8, 141-147.
Proc. Inst. Brit. Foundrymen, 1937-38, 31, 124.
Proc. Inst. Brit. Foundrymen, 1947-48, 41, Afl-75.
Russ. Cast. Prod., 1961, pp. 365-368.
Russ. Cast. Prod., 1964, pp. 83-85.
Russ. Cast. Prod., 1966, pp. 129-133.
Proc. Insf. Brit. Faundrymen, 1949, 42, A61-77.
Russ. Cast. Prod., 1967, pp. 318-320.
Russ. Cast. Prod., 1970, pp. 484-487.
 Residual Stresses in Iron Castings
Pioc. Inst. Brit. Foa~idiymen, 1958, 51, 355.

RKSS. Cast. Prod, 1962, pp. 563.

y, Gerihikov, A. M. and Rzrss. Cast. Piod., 1970, pp. 314-317.

Kotsycbinskii, 0. Y u .

y. McGoIdrick, R. T. a n d J . Am. Soc. Naval Engrs., 1943, 55 (4), 589-609

Saunders, H. E.
38. T!~irdReport of
S.~bcommittecT.S. 32
5). Behier, H. and Giesseier, 1965, 17 (4), 163-168
Tenshcff, H.K

Russ. Cast. Prod., 1965, pp. 454456.

Russ. Cast. P m h , 1967, p. 304.
4.: Bdehler, H. and Sclhepp, W. Giesserei (Tech-ivisss-Beil~fte),1961, 13 (4), 215-220.

Russ. Cast P~od., 1973, p p . 198-199.

44. Gishovich, N.G . Russ. Cust Prod., 1'163, pp. 293--294.

Russ. Casf. P~od., 1972, pp. 275-276.
16 Kotwbinskii, 0.Yu. and Rus~s.Cost. Prod., 1967, pp. 1Sj-191.
O h e m n , Y a . 1.

47. Warpov, A. Ya. et al. Russ. Carf. Prod., 1972, pp. 366-367.

Russ. Cast. Prod., 1972, pp. 413-414.

Russ. Cust. Prod., 1977, pp. 140-141.